Akshay Aradhya

Worked on a web app that allowed product managers to look at incoming client translation requests, assign translation tasks to linguists, and send the translated files back to the client. This was primarily an internal tool that improved efficiency and helped them deliver the videos on time.

As the only developer on this project, I built and designed the entire front end and API back end.

I built the front end using React, MUI 5, and Redux. It supports many neat features, such as search filters, pagination, and direct S3 uploads. The user login and all other user auth flows were taken care of by AWS Amplify and AWS Cognito.

The REST API back-end architecture was entirely serverless. It was written using Node.js and Express.js, directly hosted on AWS Lambda, and connected to API Gateway, where all requests were authenticated via AWS Cognito.

I also wrote many additional microservices and did a lot of DevOps for this project. There were many firsts for me here, which helped me learn and explore many AWS services I hadn't used before. It was an excellent learning experience, and I loved the autonomy I had.

I developed a web scraper to scrape over 100k high-quality art prints from various websites, along with rich metadata about the artwork.

The scraper was written in Python using the following libraries: selenium, beautiful soup, and requests.

The images were downloaded locally and analyzed before being uploaded to AWS S3. The image metadata was pre-processed and uploaded to DynamoDB.

Built a large-scale web scraper capable of fetching four million Twitter user details per day and per account with a single lambda function.

The architecture was entirely serverless, with a lambda function consuming events and writing back an SQS queue, so there was no upper bound limit on scaling. One could scrape 40 million users a day with ten Twitter accounts.

The scraper also fetched user, follower, and following accounts' details. This was done via an unofficial Twitter API using Requests, an HTTP library for Python. The scraper was containerized and pushed to Amazon ECR. The containerized image was directly run on AWS Lambda. Also, the scraper had CloudWatch Logs, metrics, and alarms to track and monitor its status.

The client's initial goal of 100,000 users a day was easily surpassed, and I received top-scoring feedback from the client (10/10).

Developed a beautiful SaaS platform for caterers. It was a feature-rich product, with over 200 local business users from Bengaluru signed up and using it. This was the primary SaaS product of our startup, and I was the CTO and co-founder and wrote most of the codebase powering this platform.

The front end was written in TypeScript and used React, Redux, and MUI. The back end was written in TypeScript, Node.js, and Express.js and used DynamoDB as our primary database and Elasticsearch as our querying database.

Our infrastructure was hosted on AWS, and we used various services to power this SaaS platform.

Worked on a primary post recommendation system at one of India's largest social media companies, ShareChat and the system was based on collaborative filtering and the implicit library for Python, a GPU-optimized version of collaborative filtering.

I worked on the entire system end-to-end, including the data pipeline, which ran SQL queries on BigQuery and saved post recommendations to a DynamoDB table. I made tweaks every week based on the AB testing results. In the end, we saw a minor increase in user likes and a 15-20% increase in user retention.

Built a quick and powerful search tool using Algolia to perform full-text search queries on user notes. I also created the data pipeline to transfer records from DynamoDB to Algolia using DynamoDB streams and AWS Lambda functions.
The front end for the search feature was built using CommandBar.

Developed a web-based chat application for a US client. It was a platform for support executives and customers to communicate with their team and customers. The app was very similar to what Slack is today, with some custom features.

In a team of four, I led the front-end team. My role was mainly focused on designing mockups and building apps using AngularJS and AngularJS Material. Also worked with web sockets to send messages between the server and client.

I worked on all modern browser features, such as browser desktop notifications, AngularJS Material theming, and user session management. My team received much positive feedback from the client and the management team.

A Selenium script that would scan geographically across the city and store the results for a particular search keyword directly to our DynamoDB database. It mostly gathered all the basic details provided by Google Maps. This script is highly customizable and can be used for any city and search keyword.

I was the lead developer who built the software. As my startup's CTO and lead developer, we needed to contact a few caterers and snack shops in Bangalore who would be willing to sign up for our SaaS platform. This helped us onboard another 100 business clients on our SaaS platform.

A private Grafana dashboard using Google BigQuery as the back end. I wrote custom SQL queries for most of the data displayed here.

Data was collected and uploaded by a background agent running on a GCP compute VM instance. Data was batched and uploaded using the load job functionality of BigQuery to keep the costs low and stay within BigQuery limits.